Documentation can be browsed on Read the Docs or GitHub.
For an up-to-date-version, clone this repository, use make setup to install OCaml and all dependencies, and make to build.
Alternatively: Install opam, and then do opam install goblint. Warning: The OPAM package is updated infrequently.
Run goblint: ./goblint tests/regression/04-mutex/01-simple_rc.c.
If something goes wrong, take a look at travis-ci.sh for an example Ubuntu/macOS setup.
Goblint relies on GNU cpp to preprocess source files - the default clang cpp on macOS will not work.
You can install it with brew install gcc (first do xcode-select --install if you don't want to build from source). You can check src/config.ml to see what command is used to call cpp.
For Windows, we recommend using WSL.
You can run Goblint in a Docker container using: docker run -it voglerr/goblint ./goblint --help.
A virtual machine containing Goblint can be set up using Vagrant: vagrant up && vagrant ssh.
The analysis results are printed to stdout by default.
Adding --html saves the results as XML, which is then transformed to be viewable in a web browser.
Use make jar to build the needed Java program.
./goblint --html path/to/file.c
Open result/index.xml in a browser of your choice.
Depending on the browser security settings, it might be necessary to serve the result directory from a webserver to access it.
This can be done by e.g. running python3 -m http.server in the result directory. The results should then be accessible at http://localhost:8000